Audio Recorder and Transcriber

Overview

The Voica is a desktop application that allows you to record system audio and transcribe it into text, copying the transcription directly to your clipboard. It leverages OpenAI's Whisper model for transcription and is ideal for scenarios like tech interviews, meetings, lectures, or any situation where you need to quickly capture spoken content.

Features

• Record System Audio: Capture audio from your system's output device.
• Transcribe Audio: Convert recorded audio into text using OpenAI's Whisper model.
• Clipboard Integration: Automatically copy the transcription to your clipboard for quick access.
• Custom Hotkeys: Start and stop recording using customizable keyboard shortcuts.
• Model Selection: Choose from different Whisper models (tiny, base, small, medium, large) based on your accuracy and performance needs.
• User-Friendly Interface: Simple and intuitive GUI built with PyQt6.
• Local Processing: All transcription is done locally, ensuring your audio data remains private.

Table of Contents

• Installation
  • Prerequisites
  • Setup Instructions
• Usage
  • Running the Application
  • Configuring Audio Devices
  • Recording and Transcribing
• Configuration
  • Hotkey Customization
  • Model Selection
• Troubleshooting
• Contributing
• License
• Acknowledgements

Installation

Prerequisites

• macOS: 14.5 or higher
• Python: 3.12 or higher
• Poetry: for dependency management
• Homebrew: for installing BlackHole

Setup Instructions

1. Clone the Repository

   git clone https://github.com/yourusername/audio-recorder-transcriber.git
   cd audio-recorder-transcriber

2. Install Python Dependencies

   We recommend using Poetry for dependency management.

   • Install Poetry (if not already installed):

     curl -sSL https://install.python-poetry.org | python3 -

   • Install Dependencies:

     poetry install

   Alternatively, you can use pip and requirements.txt:

   pip install -r requirements.txt

3. Install BlackHole for macOS

   BlackHole is a virtual audio driver that allows you to route audio between applications.

   • Install via Homebrew:

     brew install blackhole-2ch

   • Set Up Multi-Output Device:

     • Open Audio MIDI Setup (found in /Applications/Utilities/).
     • Click the + button at the bottom left and select Create Multi-Output Device.
     • Check both your main output device (e.g., internal speakers) and BlackHole 2ch.
     • Right-click the new multi-output device, select Use This Device For Sound Output.

4. Configure System Input

   • Open System Preferences > Sound > Input.
   • Select BlackHole 2ch as your input device.

Usage

Running the Application

Activate your virtual environment (if using Poetry):

poetry shell

Run the application:

poetry run start

Configuring Audio Devices

1. Select Input Device:

   • In the app, choose BlackHole 2ch from the input device dropdown.

2. Select Whisper Model:

   • Choose the desired Whisper model based on your needs:
     • tiny: Fastest, less accurate.
     • base: Balanced speed and accuracy.
     • small, medium, large: Increasingly accurate but require more resources.

Recording and Transcribing

1. Set Hotkey:

   • Choose a hotkey from the dropdown to start/stop recording (default is Caps Lock).

2. Start Recording:

   • Press the selected hotkey or click the Start Recording button.
   • The status label will display Recording....

3. Stop Recording:

   • Press the hotkey again or click the Stop Recording button.
   • Transcription will begin automatically.

4. Access Transcription:

   • Once transcription is complete, the text will be displayed in the log area.
   • The transcription is automatically copied to your clipboard.

Configuration

Hotkey Customization

• Change Hotkey:

  • Select your preferred hotkey from the Select Hotkey dropdown.
  • Supported hotkeys: Caps Lock, F1, F2, F3, F4, F5.

Model Selection

• Change Whisper Model:

  • Select a model from the Select Whisper Model dropdown.
  • Models vary in size and accuracy.

Troubleshooting

• No Audio Input Detected:

  • Ensure BlackHole 2ch is set as your system input device.
  • Verify that the multi-output device includes BlackHole 2ch.

• Cannot Start Recording:

  • Check that the selected input device is correct.
  • Ensure the app has permission to access the microphone (System Preferences > Security & Privacy > Privacy > Microphone).

• Transcription is Inaccurate:

  • Try selecting a larger Whisper model for better accuracy.
  • Ensure that the audio quality is good and clear.

• Hotkey Not Responding:

  • The app window must be in focus for hotkeys to work.
  • Custom global hotkeys are not supported due to macOS security restrictions.

• High CPU Usage or Slow Performance:

  • Larger Whisper models require more computational power.
  • Use a smaller model like tiny or base for better performance.

Contributing

Contributions are welcome! Please open an issue or submit a pull request on GitHub.

1. Fork the Repository

2. Create a Feature Branch

   git checkout -b feature/YourFeature

3. Commit Your Changes

   git commit -m "Add YourFeature"

4. Push to Your Fork

   git push origin feature/YourFeature

5. Open a Pull Request

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgements

• OpenAI Whisper for the speech recognition model
• BlackHole for the virtual audio driver
• PyQt6 for the GUI framework

---

Disclaimer: This app is provided "as is" without warranty of any kind. Use at your own risk.